Kruschwitz, J.D. *, List, D. *, Rubinov, M. , Walter, H.

Transcription

Kruschwitz, J.D. *, List, D. *, Rubinov, M. , Walter, H.
GraphVar: A user-friendly toolbox for comprehensive
graph analyses of functional brain connectivity.
Kruschwitz, J.D. 1,2. *, List, D. 1,2 *, Rubinov, M. 3, Walter, H. 1
* equal contribution
1Charité
Universitätsmedizin, Berlin, Germany; 2Technische Universität Dresden, Dresden, Germany; 3University of Cambridge, Cambridge, United Kingdom
INTRODUCTION
•
Graph theory provides a powerful formalism for comprehensive description of global
and local topological network properties of functional brain connectivity (Bullmore and
Sporns, 2009; Rubinov and Sporns, 2010).
•
Recently developed software packages such as the Brain Connectivity Toolbox
(Rubinov and Sporns 2010) have contributed to graph theory’s increasing popularity
for characterization of functional brain networks.
•
However, most current software packages are command-line based and may require
some programming experience. This precludes their use by many users without a
computational background, including psychologists and physicians, whose research
could otherwise benefit from graph-theoretical methods.
METHODS
•
Here we developed “GraphVar”, a user-friendly graphical-user-interface (GUI)-based
toolbox for comprehensive graph-theoretical analyses of brain connectivity, including
network construction and characterization, statistical analysis on network topological
measures, and interactive exploration of results.
•
By combining together features across multiple current toolboxes, such as the Brain
Connectivity Toolbox, the Graph Analysis Toolbox, and the Network Based Statistic
Toolbox (BCT, Rubinov and Sporns 2010; GAT, Hosseini et al., 2012; NBS, Zalesky
et al., 2010), GraphVar represents a comprehensive collection of graph analysis
routines for functional neuroimaging researchers.
SETUP AND FEATURES
WORKFLOW AND RESULTS EXPLORATION
Input:
Pipeline network
construction
(incl. null-model networks)
or
Export graph metrics
to spreadsheet
(csv/xlsx)
Statistical analyses
(partial) correlation,
t-tests, ANOVA
with clinical data
Interactive results
exploration
(incl. network based
statistics)
Node n
Time course…
Time course…
Node 2 Node 3
Time course…
Time course…
Node 1
Calculate network
topological measures
(incl. normalization)
Use raw connectivity
matrices
(generate null-model
networks)
and
Subject data
(e.g. clinical data)
• GraphVar offers an interactive viewer
that allows intuitive exploration of
statistical results as well as entails
correction methods for multiple
comparisons, including Bonferroni
correction and false discovery rate.
• Results can easily be exported
(csv/xlsx), saved, and reloaded.
• The program entails a detailed manual
that includes usage instructions and a
description of all the implemented
functions.
Figure 1: x-axis: thesholds/ y-axis: correlation/
blue line: original networks/ grey lines: random networks
•
•
GraphVar is a GUI-based toolbox which
runs MATLAB (MathWorks, Inc.) but
does not require any programming
experience from the user.
GraphVar contains most functions
included in the Brain Connectivity
Toolbox, and allows users to add
custom
functions
which
can
subsequently be accessed via the GUI.
GraphVar accepts correlation matrices
as input and can also generate
correlation
or
partial
correlation
matrices from input time series. The
user may also input demographic,
clinical and other subject specific data
(in spreadsheet format) for statistical
analyses.
CONCLUSIONS
•
• Generate correlation or partial correlation
matrices from input time series
• Pipeline construction of graph networks
• Generate subject specific null-model
networks
• Perform sub-network analyses
• Calculation, normalization, and export of
binary and weighted network measures
Fig 1: example non-parametric correlation testing with
random networks and global clustering coefficient across
densities
Fig 2: example correlation testing with local clustering
coefficient in 10 brain regions across densities
Fig 3: example results of a group comparison on the raw
correlation matrices
Fig 4: “network inspector” which may be used to identify
graph components
• Statistical analyses:
•
•
Figures results exploration:
Toolbox features:
•
Correlation and partial correlation
analyses and group comparisons
(t-test, ANOVA) on the network
measures but also on the raw
connectivity matrices (i.e., network
based statistics including identification
of graph components)
Statistical testing in parametric and nonparametric fashion (i.e., testing against
null-model networks, non-parametric
permutation testing)
Figure 2 : x-axis: nodes/ y-axis: thresholds/ colour: correlation
Figure 3 : x-axis: nodes/ y-axis: nodes/ colour: group difference
Figure 4: left: probability table/ right: graph component
for beta testing please email to: [email protected]
• GraphVar is a new comprehensive and user-friendly toolbox for easy pipeline network construction, graph
analysis, statistical analysis on network topological measures, network based statistics, and interactive
results exploration.
• The availability of such comprehensive network analysis tools may increase the accessibility of graphtheoretical connectivity to neuroimaging researchers.
REFERENCES
1
Bullmore, E., Sporns, O. (2009). 'Complex brain networks: graph theoretical analysis of structural and functional
systems', Nature reviews neuroscience, 10, 186-198.
2
Hosseini, S.M.H, Hoeft, F., Kesler, S.R. (2012) 'GAT: A Graph-Theoretical Analysis Toolbox for Analyzing BetweenGroup Differences in Large-Scale Structural and Functional Brain Networks', PLoS ONE 7(7): e40709.
doi:10.1371/journal.pone.0040709
3
Rubinov, M., Sporns, O. (2010) 'Complex network measures of brain connectivity: uses and interpretations',
Neuroimage, 52, 1059-69.
4
Zalesky, A., Fornito, A., Bullmore, E.T. (2010) 'Network-based statistic: identifying differences in brain networks',
Neuroimage, 53, 1197-207.
contact: [email protected]